The Golden Age of Open Source Applications

We are seeing a Cambrian explosion of open-source software applications.

I have written before about how AI has made software dramatically easier to build. A developer can now create in days what used to take a small team months. The cost of producing code is heading toward zero.

But distribution and managing software are still hard.

One interesting side effect is that these new dynamics are causing open source to explode.

We are seeing developers start with a paid product, put it out into the world, and discover that few people are buying it. Before they give up and throw the product away, many are open-sourcing it.

Developers are also increasingly starting with open source. They see an expensive product, realize the core functionality is now cheap to reproduce, and build a free open-source version from day one.

Look at voice-to-text dictation as an example. Wispr Flow launched a few years ago, letting people talk to their computer and have it smartly create text. Developers realized this was relatively easy to copy, and dozens of open-source projects have emerged, such as OpenWhispr, Handy, Muesli, VoiceInk, Ghost Pepper, FreeFlow, and more. Many run transcription locally. Users get lower cost, more privacy, and the ability to change how the product works.

Open source may not give developers a financial return initially, but it does have substantial benefits.

A developer can earn credibility. Stars and contributors make the work visible. A good project becomes a public audition for a job or fundraising for the next company. It can become a distribution channel for another product. Or the project itself can turn into a business through managed hosting or a paid service around the free code.

There is an emotional payoff too. You built something because you wanted people to use it. Maybe no one would pay for it, but people start installing it, starring it, contributing to it, and telling you it is useful. That feels much better than watching the code die in a private repository.

Some developers already made enough money and want their work to have an impact without starting another company. Others are trying to get hired. The motives are different, but AI makes open source an easier choice because the original investment is so much smaller.

Why Now?

Open source has obviously been around for a long time. Linux, WordPress, PostgreSQL, and thousands of other projects already proved that it can become critical infrastructure.

What has changed is the volume.

We are not just getting open-source infrastructure built over years by large communities. We are getting free open-source versions of expensive products within days or weeks of the category becoming interesting.

Look at personal agents.

In 2025, agents were largely in the realm of enterprise software, with companies raising venture capital and trying to sell agents for thousands of dollars a month.

Then in early 2026, OpenClaw emerged as an open-source personal agent and took the category by storm.

After that came dozens of other open-source projects: Hermes, NanoClaw, NemoClaw, and many more.

The same thing is happening with vibe coding. While CodeRabbit, Cursor, and Lovable took off as SaaS developer tools, you now see open-source projects like RoboRev, Superpowers, Compound Engineering, and dozens of other free tools that add capabilities to models through skills and plugins.

And if none of those do exactly what you want, you’re only a coding agent and a few hours away from creating your own skill, plugin, MCP, or full-fledged app.

Most of these open-source projects will not survive.

Hundreds of people will build roughly the same thing. A few projects will develop real momentum, and everyone else will either stop, contribute to the winners, or fork one for a specific need.

The Demand Side

What I find especially interesting is that this isn’t just a supply-side phenomenon.

Developers are producing vastly more open-source software. But there is now more demand for it too.

Historically, companies often didn’t want open source even when the software itself was free.

Someone still had to install it, figure out the dependencies, configure it, deploy it, secure it, integrate it, maintain it, and customize it. For a company, that could easily make open source more expensive than simply paying for SaaS.

AI doesn’t just make open source easier to build. It makes it easier to install and use.

You can ask Claude or Codex to install, deploy, secure, configure, and customize these projects for you.

Imagine a company considering Jira or Trello. Instead of paying for another SaaS product, it finds an open-source project that gets 80% of the way there. An AI coding agent can help deploy it, connect it to Slack and GitHub, and customize the workflows for the way that particular company works.

The open-source version doesn’t have to reproduce every feature in Jira. It just has to do what that company actually needs.

Companies are also tired of renting dozens of SaaS products that don’t quite fit, don’t talk to each other, and become more expensive every year. Open source gives them control over their data and code, less vendor lock-in, and the ability to customize the software instead of being locked into someone else’s definition of how the product should work.

So at exactly the same time that developers are creating vastly more open source, AI is making it easier for companies to actually use it.

Market Landscape

So where does this leave the software market?

As I have written previously, A lot of money will still go toward the easy button of using the existing big players.

Most companies don’t want to operate, upgrade, support, and maintain custom software, even if it’s built on open source. Great SaaS products will continue to have an enormous advantage in convenience, design, reliability, and support.

But I think a meaningful amount of money will also be made around open source.

Not necessarily by charging for every copy of the code.

Instead, people will pay for hosting, deployment, security, support, updates, customization, proprietary models, databases, enterprise management, and everything else that makes the free software dependable and easy to use.

Some enormous companies have already been built this way.

We are also going to see open source become a much more viable distribution channel (which I probably should write about more). A developer may give away the software and build the business around what the project distributes: a hosted service, a model, a database, consulting, enterprise features, or even themselves.

Limitations of Open Source

I should caveat that it is not all full steam ahead. Open source still has real limitations.

Historically, open source has not been fantastic at UI and product refinement. An open-source project can get you 70% or 80% of the way there, but paid SaaS usually puts on the finishing touches that turn a useful product into a “wow” experience.

Successful SaaS companies have dedicated product designers and engineers working every day on hundreds of small details: onboarding, interactions, edge cases, performance, design, and all the little tweaks that take a product from feeling okay to feeling great.

Open source usually doesn’t have that same level of refinement or the dedicated resources required to get there. It ends up having a bunch of rough edges.

There are other problems too. A maintainer can lose interest. Security issues can sit unfixed. Documentation gets stale. Features remain half-finished.

AI may create a new problem too. Open source used to feel like a small community building something together. Now maintainers are getting inundated with AI-generated pull requests, so the center of gravity will shift from reviewing a small number of contributions to managing an unwieldy GitHub issue queue full of people and their agents.

The category matters as well.

Backend tools and single-user products are probably the easiest fit. One technical person can tolerate some rough edges, and increasingly an agent can help fix them.

Multi-user products are harder. They need permissions, onboarding, collaboration, reliability, and a consistently great UI. Open-source products historically haven’t been particularly good at these things compared with SaaS.

AI can help but won’t magically eliminate these problems entirely.

The big unknown is whether this new explosion of open source brings enough developers, users, contributors, and eventually businesses around the winners to give them the level of refinement that historically only the best commercial software has had.

There are enormous tailwinds pushing in that direction. But turning a good open-source project into great software is still a big lift.

What This All Means

AI is making software incredibly cheap to create.

Agents are making open-source software increasingly cheap to consume.

Those two changes reinforce each other.

Developers can create and release vastly more software. Agents can search through that software, install it, configure it, fork it, and customize it for an individual person or company.

That means we are going to get an enormous library of open-source software that agents can use as building blocks.

Most of it will disappear. Some projects will become standards. Some will become the foundation for thousands of customized versions that their original developers never imagined.

And some very large businesses will be built around making all of this open-source software dependable, secure, hosted, maintained, and easy to use.

For years, the comparison was:

$30/month/user SaaS vs. free software that takes me days to configure.

Increasingly, the comparison may become:

$300/month/user SaaS vs. free software my agent installed and customized for me.

That’s a very different competitive landscape.

And I think we’re only at the beginning of it.

Note: There is a separate explosion happening in open-weight AI models from DeepSeek, Qwen, GLM, and others. That’s interesting too, but this post is about the explosion in applications.

(originally posted at https://iamcharliegraham.substack.com/p/the-golden-age-of-open-source-applications)